In the past year, Doolittle and colleagues have determined the structures of the terminal human fragment D domain (M.W. 86 kDa) and the crosslinked D-D fragment obtained from fibrin (Spraggon et al., 1997). Using molecular replacement techniques (with X-ray data collected both at the Brookhaven and at CHESS), we have succeeded in positioning the fragment D domains in the unit cells of the flash-frozen (as well as 4 C) crystals of the intact bovine fibrinogen molecule. We find that the end-to-end contacts made between symmetry-related molecules in our crystals are the same as those found by Doolittle and colleagues in the human D-D dimer. Using electron density difference maps, we clearly see coiled-coil regions connecting the terminal D and central E regions of the molecule, and can now locate the bends in these coiled coils. These bends appear to be the primary points of flexibility of the molecule. Using a variety of techniques, including improved data processing programs, refinement, and density modification, we are now extending the resolution of the electron density map to 3.5 (the present limit of the flash-frozen data sets). We are now tracing the chains in the Fragment E and connecting coiled coil regions, and are on the threshold of the first near-atomic resolution picture of virtually the whole fibrinogen molecule (Brown et al., in preparation). Using protein supplied to us by L. Medved, we have also crystallized the central Fragment E of bovine fibrinogen (which consists of a disulphide bridge-stabilized globular domain connecting two short coiled coils). Data to 2.8 resolution have been collected at CHESS from a native and 7 heavy metal-soaked Fragment E crystals. In attempting to solve this crystal structure, we are also using models of Fragment E derived from our results on the whole molecule. This information is vital for understanding the assembly of fibrinogen into the blood clot and for establishing a comprehensive rational drug design approach for treatment of clotting disorders.